“F13.4d replication: 0237's K4 statistic (5.05 at table-then-move, key length 2, above all 100 shuffles, p 0.0099) against 300 fresh climbed shuffles, the search unchanged”

1. Shuffle K4's letters 300 times.
2. Run the same search on K4 and on every shuffle.
3. Count how many shuffles score at least as high as K4; fewer than 15 of 300 would replicate the result.

Order A on a planted message: K3's rule with step 7 (letter i goes to place 7i mod 97), then the KRYPTOS table with SHADOW:

  message   THEHOTAIRESCAPINGFROMTHECHAMBERCAUSEDTHEFLAMETOFLICKERBUTPRESENTLYDETAILSOFTHEROOMWITHINEMERGEDTH
  carved    DPILCFDYKUWKWYMWSBAQVXLSDZFPDXGNGVJRECNUCIGSGRXELPEUQRLIBXISWIQUJRSJFUESQZLIISQZBUWRGUNMCQQLZRKTU
  undone    THEHOTAIRESCAPINGFROMTHECHAMBERCAUSEDTHEFLAMETOFLICKERBUTPRESENTLYDETAILSOFTHEROOMWITHINEMERGEDTH

The search tries all 27,852 moves in both orders, every length 1-20 and the 8 ways of using the table.

If K4's carved order holds a structure that a move plus a short periodic key turns into English-like text (C1(a), C11, the width-7 and width-21 fingerprints of 0056 read as a move), the excess should replicate against fresh shuffles; if it was the luck of 100 copies (the largest of 100 shuffles' statistics was 3.74, K4 5.05), it should not. Pre-registered: replicated when K4's statistic S (largest z over the 40 cells) is above all but a share p < 0.05 of 300 fresh shuffles; the B2 cell's own rank reported beside it. A replicated excess with the known letters out of place would be written up as a lead for the A2 question, not a solution.

Not replicated: K4's statistic falls inside the 300 fresh shuffles at p about 0.01-0.05 if the excess is real but small, or above 0.05 if it was luck; I expect above 0.05 against the fresh shuffles' wider spread. Zero candidates pass the gates.

1. 0237's attack unchanged except the seed (238), the shuffles (300) and no plants.
2. K4 and every shuffle solved by the same screen and climb (27,852 moves, both orders, 8 classes, lengths 1-20).
3. S = largest z over the 40 cells against the 300 shuffles; K4's p; B2's rank.

All 40 candidates, lowest loss first.

REPLICATED: K4 statistic 4.62 at B2, p 0.023 against 300 shuffles; plant power ; decided lengths []

What was searched, as the harness recorded it.

item

F13.4d

search reads known letters

no

moves

27,852

orders

A move-then-table, B table-then-move

classes

std/std/vig, std/std/beau, std/kry/vig, std/kry/beau, kry/std/vig, kry/std/beau, kry/kry/vig, kry/kry/beau

key lengths

1, 20

screen

per-column letter-frequency shifts, every move scored by quadgrams; 30 best per order and length

climb

coordinate ascent, move held, screen's shifts and 5 random starts; MIX8m rescoring

texts

301

k4 copies

300

k4 statistic

4.623

k4 best cell

B2

k4 p

0.023

copies statistic max

5.385

copies statistic median

2.309

k4 rank per cell copies at or above

A1

18

A2

197

A3

292

A4

64

A5

89

A6

212

A7

151

A8

243

A9

96

A10

154

A11

242

A12

220

A13

157

A14

49

A15

88

A16

185

A17

116

A18

228

A19

134

A20

186

B1

18

B2

1

B3

239

B4

24

B5

174

B6

114

B7

173

B8

118

B9

232

B10

56

B11

22

B12

82

B13

257

B14

125

B15

10

B16

137

B17

115

B18

29

B19

40

B20

189

replicated

yes

b2 copies at or above

1

replicating

0237 (S 5.05 at B2, p 0.0099 against 100)

plant power

plant exact recovery

decided lengths at 6 of 8

none

k4 best reading

cell

B2

reading

NBENDJZERANDBCKNFILLBAADWRONNWIIHLGARTNRHHKORCULLHCNCSIDEPXMEDFCKRHALORKYXQTRGROWDOJOWSMSHDVNLYAT

move

two turns cw 70,19 inverse

f6 1

blocked: 0236 found one free alphabet cannot be recovered from the ciphertext alone beyond length 5 with no move

expected total time

541 texts x about 25 s = about 4 core-hours

exhaustive

no

compute seconds

151